Recently, technology for extracting correlation data from large-scale data called big data attracts attention and an increase of a data amount and improvement of a processing ability of an information apparatus are advanced. At this time, a communication speed inside an apparatus becoming a bottleneck of large-scale data processing and a communication speed outside the apparatus are also improved. For example, in communication outside the apparatus, correspondence to a standard such as 100G Ether and 32G FC is advanced and a speed increase to communication speeds of 25 Gbps and 28 Gbps per lane is advanced.
As the communication speed increases, an influence of reflection at a short distance is actualized and a frequency characteristic of a communication path does not become linear. In addition, a time of a unit interval (1UI) becomes short and an allowed amount of jitter decreases. For this reason, waveform equalization is performed using a feed forward equalizer (FFE) or a continuous time linear equalizer (CTLE) and the jitter by inter-symbol interference (ISI) is improved. However, the jitter is not compensated for. Therefore, a decision feedback equalizer (DFE) is often used to improve the ISI jitter (refer to NPL 1).
The DFE has a configuration of a digital filter and includes a plurality of taps obtained by multiplying filter coefficients by data, which are added to an input. The DFE can cancel an influence of a previous signal or an influence of reflection arriving late from a distance as much as the DFE increases the number of taps. If a frequency becomes high, a wavelength becomes short. For this reason, it is necessary to increase the number of taps of the DFE to cancel near reflection. Therefore, when the frequency increases, the number of taps of the DFE tends to increase.